Perplexity 发布了两个新的晚期交互嵌入模型 pplx-embed-v2-late，它们可以在共享的嵌入空间中检索文本、图像和页面。这些模型达到了前沿性能，并且可以在 Hugging Face 上使用。9B 模型在代理搜索任务上表现强劲，在 BrowseComp+ 问题上优于 ColBERT 模型，并在 Q2D-Web 基准测试中取得了高 Recall@1000 分数。此外，Perplexity 强调，这些模型无需 OCR 即可处理 PDF、幻灯片和扫描件，并保留了布局和表格。 AI
影响 这些新的嵌入模型可以增强搜索能力和跨模态查询，可能为检索系统设定新的基准。
排序理由 Perplexity 是一个发布新嵌入模型的前沿实验室。
AI 生成摘要 · Google Gemini · 来自 8 个来源。 我们如何撰写摘要 →
完整方法见我们的编辑标准。
Both models are on Hugging Face: https://t.co/JHy1DhFaCT
On MADQA, 500 questions over 800 PDFs, it reaches 92.4%, the top result among retrievers. https://t.co/uxW8LDau2c
In agentic search, a GPT-OSS-120B agent using the 9B model answers 64.0% of BrowseComp+ questions correctly, 4.9 points ahead of the next ColBERT model, with fewer searches than any baseline. https://t.co/hFngH5ES7D
On Q2D-Web, about 70,000 production-derived search queries over 190M web documents, the 9B and 0.6B models reach 74.8% and 73.6% Recall@1000, against 69.3% for nemotron-embed-8b. https://t.co/FwltQP6jGN
Both sizes are distilled token by token from one 18B teacher, so they share one embedding space. A corpus indexed with the 9B model can be searched with 0.6B queries. That lifts ViDoRe v3 from 62.3% to 63.5% with no added query cost. https://t.co/SglIFldtGQ
The models also embed images and rendered pages directly, so PDFs, slides, and scans are searched without OCR, keeping tables, figures, and layout that text extraction drops. https://t.co/iZmlOKG1ye
Dense embedding models compress each document into one vector, which loses detail as pages get longer or more visual. pplx-embed-v2-late keeps a 128-dimensional vector per token and scores with MaxSim, so each query token is matched to its closest token in the document. https://…
We're releasing pplx-embed-v2-late, two late-interaction embedding models that retrieve text, images, and pages with a shared embedding space for cross-model querying. Both models achieve frontier performance and are publicly available on Hugging Face. https://t.co/AM7YEyQkKE h…